>>I have an index routine that uses a KEYS.DBF file. This file contains the filename, index name, key expression and Ascending/Descending indicator.
>>
>>Is there a way to populate this file by reading the .dbc directly?
>>
>>Ideally, I'd like to run "LoadKeys()" at app startup as well as intermittently during development.
>>
>>TIA
>Larry,
>BTW VFP compared to FP2.x also has primary(), candidate() etc. + more enhanced functions like "alter table ... foreign key ...".
>Also a shortcut to all this is to generate programatic form of DBC with gendbc or gendbcx (www.stevearnott.com) and use it.
>Cetin

Thank you all, I didn't expect so many ideas :)

Here's what I finally used

CLOS ALL
CLEA ALL
CLEA

SET TALK OFF
SET SAFE OFF
SET ESCA ON

USE KEYS EXCL
ZAP

SELECT 0
CREATE TABLE IndexLst (FileName c(10), IndexName c(10) )

SELECT * FROM main.DBC WHERE objecttype="Table" INTO CURSOR Mytables
SCAN
    IF OBJECTNAME<>"keys"
        SELECT * FROM main.DBC WHERE objecttype="Index" AND ParentId = Mytables.ObjectID INTO CURSOR MyIndexes
        SCAN
            INSERT INTO IndexLst ( FileName, IndexName ) VALUES ( Mytables.OBJECTNAME, MyIndexes.OBJECTNAME )
        ENDSCAN
    ENDIF
ENDSCAN

CLOS ALL
CLEA ALL
CLEA

USE IndexLst
SCAN
    SELECT 0
    USE ( IndexLst.FileName )
    SET ORDER TO ( IndexLst.IndexName )
    cIndexExpr = KEY()
    ? IndexLst.FileName, IndexLst.IndexName, cIndexExpr
    INSERT INTO KEYS ( FLNAME, TAGNAME, INDEXEXPR ) VALUES ( UPPER( IndexLst.FileName ), UPPER( IndexLst.IndexName ), cIndexExpr )
    USE
ENDSCAN

I can now use KEYS.DBF to recreate the indexes easily.

I'm a real supporter of the KISS concept :) My indexing schemes are always confined to INDEX ON ??? TAG ??? commands... no descending, unique, candidate, primary, etc... qualifiers.

Thanks again for all the feedback!